• Journal of Intelligence and Information Systems
• /
• v.16 no.4
• /
• pp.99-112
• /
• 2010

Ensemble learning is a method for improving the performance of classification and prediction algorithms. It is a method for finding a highly accurateclassifier on the training set by constructing and combining an ensemble of weak classifiers, each of which needs only to be moderately accurate on the training set. Ensemble learning has received considerable attention from machine learning and artificial intelligence fields because of its remarkable performance improvement and flexible integration with the traditional learning algorithms such as decision tree (DT), neural networks (NN), and SVM, etc. In those researches, all of DT ensemble studies have demonstrated impressive improvements in the generalization behavior of DT, while NN and SVM ensemble studies have not shown remarkable performance as shown in DT ensembles. Recently, several works have reported that the performance of ensemble can be degraded where multiple classifiers of an ensemble are highly correlated with, and thereby result in multicollinearity problem, which leads to performance degradation of the ensemble. They have also proposed the differentiated learning strategies to cope with performance degradation problem. Hansen and Salamon (1990) insisted that it is necessary and sufficient for the performance enhancement of an ensemble that the ensemble should contain diverse classifiers. Breiman (1996) explored that ensemble learning can increase the performance of unstable learning algorithms, but does not show remarkable performance improvement on stable learning algorithms. Unstable learning algorithms such as decision tree learners are sensitive to the change of the training data, and thus small changes in the training data can yield large changes in the generated classifiers. Therefore, ensemble with unstable learning algorithms can guarantee some diversity among the classifiers. To the contrary, stable learning algorithms such as NN and SVM generate similar classifiers in spite of small changes of the training data, and thus the correlation among the resulting classifiers is very high. This high correlation results in multicollinearity problem, which leads to performance degradation of the ensemble. Kim,s work (2009) showedthe performance comparison in bankruptcy prediction on Korea firms using tradition prediction algorithms such as NN, DT, and SVM. It reports that stable learning algorithms such as NN and SVM have higher predictability than the unstable DT. Meanwhile, with respect to their ensemble learning, DT ensemble shows the more improved performance than NN and SVM ensemble. Further analysis with variance inflation factor (VIF) analysis empirically proves that performance degradation of ensemble is due to multicollinearity problem. It also proposes that optimization of ensemble is needed to cope with such a problem. This paper proposes a hybrid system for coverage optimization of NN ensemble (CO-NN) in order to improve the performance of NN ensemble. Coverage optimization is a technique of choosing a sub-ensemble from an original ensemble to guarantee the diversity of classifiers in coverage optimization process. CO-NN uses GA which has been widely used for various optimization problems to deal with the coverage optimization problem. The GA chromosomes for the coverage optimization are encoded into binary strings, each bit of which indicates individual classifier. The fitness function is defined as maximization of error reduction and a constraint of variance inflation factor (VIF), which is one of the generally used methods to measure multicollinearity, is added to insure the diversity of classifiers by removing high correlation among the classifiers. We use Microsoft Excel and the GAs software package called Evolver. Experiments on company failure prediction have shown that CO-NN is effectively applied in the stable performance enhancement of NNensembles through the choice of classifiers by considering the correlations of the ensemble. The classifiers which have the potential multicollinearity problem are removed by the coverage optimization process of CO-NN and thereby CO-NN has shown higher performance than a single NN classifier and NN ensemble at 1% significance level, and DT ensemble at 5% significance level. However, there remain further research issues. First, decision optimization process to find optimal combination function should be considered in further research. Secondly, various learning strategies to deal with data noise should be introduced in more advanced further researches in the future.

• Journal of the Korean Institute of Landscape Architecture
• /
• v.38 no.3
• /
• pp.33-40
• /
• 2010

Many trails have been built and utilized in each sub-basin of Apsan Park. This study focused on the conditions and the present flora of several overused trails within three basins called Kun-Gol, Anjiroung-Gol, and Meaja-Gol in Apsan Park. There are two purposes for this study. The first is to make a plan for restoration by classifying the type of damage to these overused trails. The second is to provide preliminary data for reestablishing a more pleasant urban nature park. As a result of the survey, the average length of these trails is 9.0km and mean width is 2.0 to 2.6m. 2,108m of the sections studied(23.4% of the total) had exposed root, 3,199m of the sections studied (35.5% of the total) had exposed rock, 3,270m of the sections studied(36.6% of the total) displayed damaged pathways, and 4,841m of the sections studied (53.8% of the total) had higher soil hardness. Eighty-three areas of deterioration were found and there has been extensive property destruction. As an indication of human disturbances, the Naturalization Index and Urbanization Index appeared as: Kun-Gol, 7.0%, 5.5%; Anjiroung-Gol, 8.3%, 4.5%; Meaja-Gol 8.6%, 6.2%; in total, 7.8%, 8.3%. As a method of checking the level and extent of the damage process, an impact rating classification was used at each point on the trails. Kun-Gol had the highest rating and Anjiroung-Gol had the lowest rating in impact rating classification. With the impact rating classification, it was found that each trail would continue to worsen, meaning that roots and rocks are more likely to be further exposed and the trail width will continue to widen through overuse. An actual plan for reparations and restoration of the trails is needed through further study.

• KSBB Journal
• /
• v.24 no.4
• /
• pp.377-382
• /
• 2009

The culture condition of growing Chlorella minutissima was optimized to produce biodiesel for fed-batch cultivation. First, under heterotrophic cultivation, the optimum level of glucose was determined to be 10 g/L for 20 days. After, three cultivation conditions were operated: autotrophic, heterotrophic, and mixotrophic growth. The lipid level and the maximum cell concentration from the fed-batch heterotrophic process were 32.0 (%, v/v) and 15.0 (g-dry wt./L) in 20 L flask, respectively. In addition, since the relatively constant specific lipid production rate was observed as 0.040 (% lipid/g-dry wt./day) at the latter period of cultivation time, the fed-batch process could maintain continuous lipid production. Fed-batch process is higher than those values from the batch process. The lipids from the fed-batch process contained over 38% of $C_{18}$, known as the suitable composition for the biodiesel application. For mixotrophic and heterotrophic growth under fed-batch condition, glucose was proved to be an appropriate carbon source for a large scale outdoor cultivation. For fed-batch cultivation, the feeding rate of seawater medium containing glucose was decided to be 0.5 L/day. The mixotrophic cultivation maintained maximum cell concentration of 24 (g-dry wt./L) and the lipid level of 43 (%, w/w). The lipid composition from this process was also proved to be suitable for the biodiesel production. The fatty acids from the mixotrophic growth contains 18% of $C_{17}$ and 49% of $C_{18}$, implying It also tells that C. minutissima is a suitable resource of biodiesel. Especially, the mixotrophic cultivation with fed-batch process might be useful for the large scale cultivation for the biodiesel production.

• Journal of Life Science
• /
• v.19 no.7
• /
• pp.994-1002
• /
• 2009

A bacteriocin-producing bacterium identified as Bacillus licheniformis was isolated from soybean sauce. Antibacterial activity was confirmed by paper disc diffusion method, using Micrococcus luteus as a test organism. The bacteriocin also showed antibacterial activities against Bacillus sphaericus, Lactobacillus bulgaricus, Lactobacillus planiarum, Paenibacillus polymyxa, and Pediococcus dextrinicus. Optimal culture conditions for the production of bacteriocin was attained by growing the cells in an MRS medium at a pH of 6.5~ 7.0 and a temperature of 37$^\circ$C for 36$\sim$48 hr. Solvents such as chloroform, ethanol, acetone, and acetonitrile had little effect on bacteriocin activity. However, about 50% of bacteriocin activity diminished with treatment of methanol and isopropanol at the final concentration of 50% at 25$^\circ$C for 1 hr. It was stable against a pH variation range from 3.0 and 7.0, but the activity reduced to 50% at a pH range from 9.0 to 11.0. It's activity was not affected by heat treatment at 100$^\circ$C for 30 min and 50% of activity was retained after heat treatment at 100$^\circ$C for 60 min, showing high thermostability. The bacteriocin was purified to a homogeneity through ammonium sulfate precipitation, SP-Sepharose ion-exchange chromatography, and reverse-phase high-performance liquid chromatography (HPLC). The entire purification protocol led to a 75-fold increase in specific activity and a 13.5% yield of bacteriocin activity. The molecular weight of purified bacteriocin was estimated to be about 2.5 kDa by tricine-SDS-PAGE.

• 대한공업교육학회지
• /
• v.33 no.1
• /
• pp.149-168
• /
• 2008

The purpose of this study is to verify the effect of PSpice instruction on academic achievement in 'Combination logic circuit' unit of 'Digital Logic Circuit' in industrial high school. Three kinds of null hypotheses were formulated. Two classes of the third grade of C technical high school in Gyeong-buk were divided into experimental group and control group in order to verify null hypotheses. In the experimental design, 'Non-equivalent control group pretest-posttest' model was utilized. This experiment was conducted for six classes, the experimental group was applied to PSpice instruction method before the circuit traning while the control group was applied to traditional lecture oriented method before the circuit traning. Window SPSS 10.0 korean language version program was used for the data analysis and independent sample t-test was used to identify the average of each group. Significance level was set to .05 level. The results obtained in this study were as follows; First, PSpice instruction had not an effect on academic achievement according to a group type. However, these instruction had an effect on the following sub-domains; the psychomotor domain. Second, PSpice instruction had not an effect on academic achievement according to a studies level. However, these instruction for middle and low level students had an effect on the cognitive and psychomotor domain, and for middle level students had an effect on the affective domain. Third, PSpice instruction had not an effect on shortening of a training requirement. However, this instruction for low level students had an effect on shortening of a training requirement. The study results of simulation instruction was chiefly efficient in the psychomotor domain. We could know that simulation instruction is efficient as went to a low level students than an upper level students. Thus, We may make the study effectiveness in various instruction method.

• Radiation Oncology Journal
• /
• v.9 no.2
• /
• pp.185-195
• /
• 1991

Seventy four patients with pituitary adenoma received radiation therapy (RT) on the pituitary area using 6 MV linear accelerator during the past 7 years at the Division of Radiation Therapy, Kangnam St. Mary's Hospital, Catholic University Medical College. Thirty nine were men and 35 were women. The age ranged from 7 to 65 years with the mean being 37 years. Sixty five ($88\%$) patients were treated postoperatively and 9 ($12\%$) primary RT, To evaluate the effects of RT, we analyzed the series of endocrinologic studies with prolactin (PRL), growth hormone (GH), adrenocorticotrophic hormone (ACTH), leuteinizing hormone (LH), follicular stimulating hormone (FSH) and thyroid stimulating hormone (TSH) etc after RT. All but one with Nelson's syndrome showed abnormal neuroradiologic changes in the sella turcica with invasive tumor mass around supra- and/or parasella area. The patients were classified as 23 ($29\%$) prolactinomas and 20 ($26\%$) growth hormone (GH) secreting tumors, and 6 ($8\%$ ACTH secreting ones consisting of 4 Cushing's disease and 2 Nelson's syndrome. Twentynine ($37\%$) had nonfunctioning tumor and four ($5\%$) of those secreting pituitary tumors were mixed PRL-GH secreting tumors. The hormonal level in 15 ($65\%$) of 23 PRL and 3 ($15\%$) of 20 GH secreting tumors returned to normal by 2 to 3 years after RT, but five PRL and five GH secreting tumors showed high hormonal level requiring bromocriptine medication. Endocrinologic insufficiency developed by 3 years after RT in 5 of 7 panhypopituitarisms, 4 of seven hypothyroidisms and one of two hypogonadisms, respectively. Fifteen ($20\%$) patients were lost to follow up after RT.

• Radiation Oncology Journal
• /
• v.13 no.3
• /
• pp.243-252
• /
• 1995

Purpose : Intraluminal high dose rate brachytherapy is an accepted treatment for the tumors of GI tract. However, there is only some limited clinical data for intraluminal high dose rate brachytherapy for the tumors of GI tract. Materials and Methods : Between February 1991 and July 1993, 18 Patients who have the tumors of GI tract (esophageal cancer-8 cases, rectal cancer-10 cases) were treated with high dose rate Iridium-192 afterloading system (Microselectron-HDR, Nucletron CO, Netherland) at the department of therapeutic radiology, St. Mary's hospital, Catholic university medical college. Age range was 47-87 years with a mean a9e 71 years. All patients were treated with intraluminal high dose rate brachytherapy within two weeks after conventional external radiation therapy and received 3-5 Gy/fraction 3-4 times per week to a total dose 12-20 Gy (mean 17 Gy). Standard fractionation and conventional dose were delivered for external radiation therapy. Total dose of external radiation therapy ranged 41.4-59.4 Gy (mean 49.6 Gy). Median follow up was 19 months Results : The analysis was based on 18 patients, The complete response and partial response in esophageal cancer was similar (38%). Two year rates for survival and median survival were 13% and 10 months, respectively. Among 10 patients of rectal cancers, partial response was obtained in 6 patients (60%). There was no complete response in the patients with rectal cancer, but good palliative results were achieved in all patients. Conclusion : Although the number of patients was not large and the follow-up period was relatively short, these findings suggested that intraluminal high dose rate brachytherapy could be useful in the treatment of the patients with advanced tumors of GI tract.

• Journal of Marine Bioscience and Biotechnology
• /
• v.1 no.2
• /
• pp.91-97
• /
• 2006

For efficient removal of nitrogenous compounds produced in recirculating aquaculture system, the N removal characteristics of immobilized mixed microorganisms were investigated at various mixing ratios of photosynthetic bacteria (PSB) immobilized in PVA beads or CTA cubes and ammonium utilizing bacteria (AUB) immobilized in PVA beads. On the optimal medium of AUB, the maxium gas production rate was obtained at the mixing ratio of 10:40 (PSB:AUB), and the gas production rate increased as the portion of AUB beads in the mixed beads increased. When the mixing ratios of PSB:AUB beads were 50:0, 40:10, 25:25 and 10:40, the final pHs were measured to be 6.29, 6.01, 5.69 and 5.13, respectively. On the optimal medium of PSB, however, the volume and the rate of gas production decreased remarkably as the portion of AUB beads in the mixed beads increased. The final pH was measured to be approximately 6.5, regardless of the mixing ratio. In the reactions by the mixed culture of PSB cubes and AUB beads, all results showed the same tendency of those by the mixed culture of PSB and AUB beads, but the volume and the rate of gas production decreased remarkably, even with 0.2ml of gas production in control. From all the results, the use of mixed PSB and AUB beads at the ratio of 10:40 seems to be efficient to remove nitrogenous compounds in wastewater from recirculating aquaculture system.

• Journal of Food Hygiene and Safety
• /
• v.29 no.4
• /
• pp.268-277
• /
• 2014

This study assessed microbiological hazards at postharvest stage of dropwort farms (A, B, C, D, E, F, G, H, I) located in 4 different areas in Korea. The samples were assessed for sanitary indication bacteria (total aerobic bacteria, coliform, and Escherichia coli) and pathogenic bacteria (Escherichia coli O157:H7, Listeria monocytogenes, Staphylococcus aureus and Bacillus cereus). Total aerobic bacteria and coliform in 9 dropwort farms were detected at the levels of 0~7.00 and 0~4.25 log CFU/g, mL, of $100cm^2$. In particular, microbial contamination in worker's hand showed higher than cultivation environment factors. Escherichia coli was detected in several farms of soil, irrigation water, washing water and worker's hand and also, dropwort in these farms was contaminated with E. coli (positive reaction). In case of pathogenic bacteria, B. cereus was detected at the highest levels in soil. S. aureus was detected qualitatively from only one sample of dropwort washed by water. E. coli O157:H7 and L. monocytogenes were not detected. Although dropwort pass through 2 process (trimming and washing), the microbial contamination was not differ significantly before and after which indicates that current washing system was not effect on reduction of microorganism. From these results, the postharvest environment and workers have been considered as cross-contamination factors. Thus, processing equipments and personal hygiene should be managed to reduce the microbial contamination of dropwort. Accordingly management system such as good agricultural practices (GAP) criteria is needed for the safety of dropwort

• Journal of Nutrition and Health
• /
• v.52 no.6
• /
• pp.515-528
• /
• 2019

Purpose: This study examined the immunological activity and optimized the mixture conditions of Sargassum horneri (S. horneri) extracts in vitro and in vivo models. Methods: S. horneri was extracted using three different methods: hot water extraction (HWE), 50% ethanol extraction (EE), and supercritical fluid extraction (SFE). Splenocyte proliferation and cytokine production (Interleukin-2 and Interferon-γ) were measured using a WST-1 assay and enzyme-linked immunosorbent assay, respectively. The levels of nitric oxide and T cell activation production were measured using a Griess assay and flow cytometry, respectively. The natural killer (NK) cell activity was determined using an EZ-LDH kit. Results: Among the three different types of extracts, HWE showed the highest levels of splenocyte proliferation and cytokine production in vitro. In the animal model, three different types of extracts were administrated for 14 days (once/day) at 50 and 100 mg/kg body weight. HWE and SFE showed a high level of splenocyte proliferation and cytokine production in the with and without mitogen-treated groups, whereas EE administration did not induce the splenocyte activation. When RAW264.7 macrophage cells were treated with different mixtures (HWE with 5, 10, 15, 20% of SFE) to determine the optimal mixture ratio of HWE and SFE, the levels of nitric oxide and cytokine production increased strongly in the HWE with 5% and 10% of SFE containing group. In the animal model, HWE with 5% and 10% of SFE mixture administration increased the levels of splenocyte proliferation, cytokine production, and activated CD4⁺ cell population significantly, with the highest level observed in the HWE with 5% of SFE group. Moreover, the NK cell activity was increased significantly in the HWE with 5% of SFE mixture-treated group compared to the control group. Conclusion: The optimal mixture condition of S. horneri with immune-enhancing activity is the HWE with 5% of SFE mixture. These results confirmed that the extracts of S. horneri and its mixtures are potential candidate materials for immune enhancement.